PROJECT SUMMARY The pathophysiology of depression remains poorly understood, perhaps explaining why half of depressed patients do not adequately respond to available treatments, leading to chronic depression with its associated morbidity, premature mortality, and societal costs. Inflammation in the body is communicated to the brain, leading to depression in some individuals. But the brain also regulates the immune system and yet little is known about how depression-linked neural abnormalities affect systemic immunity. Without more detailed knowledge, intervening therapeutically in the dynamic, homeostatic interplay between the brain and immune system is challenging. We use real-time neurofeedback (NF) to experimentally modulate a key neural node (i.e. amygdala and hippocampus of the medial temporal lobe, MTL) that has been shown to regulate immune function pre-clinically. By comparing depressed subjects who can versus those who cannot regain plasticity or function of the MTL due to NF modulation, we take a step towards identifying immune markers of treatment response and prognosis (aim 1). Conversely, by characterizing the immune changes that occur in subjects who are able to regain MTL plasticity/function, biomarkers of therapeutic brain changes can be identified (aim 2). Moreover, if this NF-induced change in immunity plays a mechanistic role in recovery from depression, then medications can be developed to augment this therapeutic immune response. The NF paradigm used here is focused on the amygdala. In two clinical trials, we showed that training depressed subjects to upregulate their left amygdala activity while recalling positive autobiographical memories significantly reduced symptoms and increased fMRI functional connectivity of the amygdala and hippocampus relative to depressed subjects in the sham group. Here, the main hypotheses are that: (1) baseline levels of inflammatory cytokines and neurotoxic kynurenine metabolites will be higher in subjects who are unable to effectively upregulate their amygdala (non- responders) compared with subjects who effectively upregulate their amygdala (responders). (2) Compared with non-responders, responders to NF will show increases in anti-inflammatory cytokines and neuroprotective kynurenines post-NF training. To test these hypotheses, we will measure depressive symptoms, circuits (amygdala activity and changes in MTL connectivity), and molecules (cytokines and kynurenines) using a within-subjects design. Unmedicated subjects with depression (PHQ-9 score ?10, n=32) will complete two, one hour-long sessions (V1-V2) of NF (no sham). V1 and V2 will be held two weeks apart and will incorporate both training and evaluative components. Serum will be obtained at V1 and V2 and at a third visit without fMRI at week 4 (V3). This research is highly impactful because we need to understand the mechanisms underpinning the bidirectional relationship between inflammation and depression in order to facilitate development of new pharmacological or behavioral therapies, and identify predictive, prognostic, and monitoring biomarkers.